A GRASP Metaheuristic for the Coverage of Grid Environments with Limited-Footprint Tools

Coverage of known environments is a task involved in several applications of autonomous mobile robots, like patrolling, search and rescue, and cleaning. The (single robot) coverage problem can be formulated as that of finding the optimal tour that, when followed, allows a robot to cover with its tool (e.g., a sensor or a brush) all the points of the free space of a given environment. Most of the current methods for coverage discretize the environment in cells, possibly of different shapes. In this paper, we consider a setting in which the environment is represented as a grid of equal square cells and in which a robot has a tool with limited range and angular field of view, able to cover a set of cells from a given pose. We propose an efficient covering method based on a on Greedy Randomized Adaptive Search Procedure (GRASP) metaheuristic approach that iteratively constructs a feasible solution and tries to improve it through local search. Results of experimental activities show that the proposed method produces solutions of better quality than those of a state-of-the-art method, in an efficient way.